In this paper we present a procedure for fabricating an array of micropolarization filter array via an optimized interference lithography and microfabrication procedure. The filter array is composed of two linear polarization filters offset by 45 degrees with pixel pitch of 18 microns. The individual polarization filters are composed of aluminum nanowires with 140 nm pitch, 140 nm height and 70 nm width. The maximum extinction ratio of the pixelated filters is measured to be 95 at 700nm wavelength.
©2011 Optical Society of America
Advancements in nanofabrication technology and nanofabrication techniques [1–3] have enabled the inception of division of focal plane polarization imaging sensors for the visible spectrum [4–9]. Division of focal plane polarimeters monolithically integrate array of imaging elements with pixelated polarization filters on the same substrate and are capable of recording the optical properties of partially polarized light at every imaging frame. The idea of monolithic integration of optical filters with high resolution imaging arrays can be traced back to the Bayer color filter pattern and the down of the solid state color sensors era . The monolithic integration of color filters with CCD and CMOS imaging arrays yielded compact and robust color sensors which have been perfected and proliferated for the last 30 years.
The challenges in developing division of focal plane polarization sensors for the visible spectrum have been in the reliable fabrication of large areas of pixelated optical filters capable of filtering the linear polarization properties of the incoming light wave and integrating these filters monolithically with CMOS or CCD imaging arrays. The pixelated polarization filters have to contain an array of optical filters whose pitch is matched to the one of the underlying imaging sensor. Furthermore, these filters have to be composed of periodic metallic wires which will filter the polarization properties of the incoming light wave. The width of the metallic wires in the pixelated polarization filters should be around ~1/5th of the incident wavelength . For example, imaging polarization properties of infrared wavelengths between 3 μm and 5 μm has been achieved via wire grid linear polarizers with 475nm wide periodic metallic wires . These periodic wires can be fabricated via regular photolithography procedure using an I-line aligner operating at 375nm . In order to image wavelengths in the visible spectrum down to 450 nm, the metallic wires in a pixelated linear polarization filter should have a width of ~90 nm or less. Fabricating feature sizes bellow 200 nm is not possible via standard UV lithography and advance nanofabrication techniques should be employed in order to realize these filters.
The purpose of the micropolarization array is to filter the incoming light wave prior to the absorption by the silicon photodiode with sever different linear polarization filters. A successful candidate for a micropolarization array has to satisfy four basic criteria. First, the array has to be able to filter the incoming light wave with at least two non-orthogonal filters as well pass an unfiltered version of the light wave. Second, the pixel-pitch of the micropolarization filter has to match the pitch of the imaging sensor. Third, the extinction ratios of the polarization filters i.e. the ratio of the parallel polarized light to cross polarized light has to be high in order to be able to extract useful information .
In this paper, we describe a procedure for fabricating a dual layer aluminum nanowire polarization filter suitable for division of focal plane polarization imaging sensors for the visible spectrum. The dual layer micropolarization array is fabricated using combination of optimized interference lithography and microfabrication steps resulting in a pixelated polarization filter array with 18 micron pitch and extinction ratios of 95 at 700nm incident light wavelength. The filter array is abutted to the surface of a custom CMOS imaging sensor and optical test are performed on the array.
The rest of the paper is organized as follows. In Section 2 we provide a theoretical framework for designing pixelated polarization filter arrays for division of focal plane polarization sensors. A detail description of the microfabrication procedure for designing dual layer polarization filter array is presented in Section 3. Section 4 presents experimental results from the dual layer filter and concluding remarks are presented in Section 5.
2. Overview of polarization properties
In order to compute the optical properties of partially polarized light, the first three Stokes parameters should be captured using a combination of imaging arrays and linear polarization filters. There are different ways of computing Stokes parameters and one of them is presented by Eqs. (1) through (3):
In Eqs. (1) through (3), It is the total intensity of the incoming light wave; I(0°) is the intensity of the incoming light wave after being filtered with a 0° linear polarization filter; I(45°) is the intensity of the incoming light wave after being filtered with a 45° linear polarization filter. Hence, in order to compute the first three Stokes parameters, the incoming light wave should be filtered with two linear polarization filters offset by 45 degrees as well as the intensity of the unfiltered light intensity should be recorded by an array of photo detectors.
A block diagram of a division of focal plane polarization sensors based on guidelines established from Eqs. (1) through (3) is presented in Fig. 1 . The micropolarization filter array is composed of two distinct pixelated linear polarization filters offset by 45° and one pixel without any linear polarization filters. The pattern of pixelated polarization filters is repeated across the entire imaging array. Typically, polarization information is computed on a block of 2 by 2 pixels, which is also known as super pixel. There are other possibilities for designing micropolarization filter arrays and depending on the filter array pattern the signal to noise ratio of the computed Stokes parameters will be effected [5, 13].
3. Microfabrication procedure for dual layer nanowire filter array
Fabricating nanowires polarization filters are not possible using standard lithography process due to diffraction limitations of ultraviolet (UV) lithography. Since polarization filters are composed of periodic structures, interference lithography or holographic lithography is suitable for fabricating these filters [1, 2, 14]. Holographic lithography is realized by superposition of two coherent electromagnetic waves resulting in a periodic spatial pattern with low and high intensity values. The period of the interference pattern generated by the two waves can be estimated to be X = λ/2sin(Θ/2), where λ is the wavelength and Θ is the interference angle of the two waves. Furthermore, the wavelength of the two electromagnetic waves has to be compatible with the exposure wavelength of the photoresist in order to be able to pattern the photoresist. Most photoresists are sensitive in the UV region ranging from 200 nm up to ~400 nm.
In our experimental setup, we used a solid-state neodymium-doped yttrium aluminum garnet 532 nm laser together with a frequency doubler in order to generate an electromagnetic wave with 266nm wavelength. Using a combination of beam splitters and mirrors, two waves are generated from the source and are aligned to interfere at ~120 degrees. The fringe pattern generated by the two interfering waves has periodicity of 140nm over a circular area with a diameter of ~3 cm.
The detail fabrication procedure for designing a dual layer nanowires polarization filters is depicted in Fig. 2 and described below:
- (1) The starting substrate for the filter array is a microscope glass slide. The glass slide is first planerized via a chemical mechanical polishing (CMP) procedure optimized for glass substrates. The variations of the glass surface within an area of 3 cm by 3 cm are +/−5 nm. The surface of the glass slide is cleaned by soaking the sample in isopropyl alcohol and acetone separately. The glass sample is then placed in a reactive ion etching tool and the surface is further cleaned with oxygen via reactive ion etching (RIE) power of 300 W for 20 minutes.
- (2) The glass surface is coated with 140 nm of aluminum and 20 nm of SiO2 via e-beam deposition and chemical vapor deposition respectively. The SiO2 will be used as a hard mask to etch the aluminum layer.
- (3) Thin layer of photoresist is spin coated on the sample. A diluted S1805 photoresist is spin coated at 4500 rpm for 45 seconds and ~150 nm thick photoresist is created on the surface of the sample. The sample is then baked at 90° C for one minute (Fig. 2a).
- (4) The interference pattern generated by the neodymium-doped yttrium aluminum garnet laser is used to pattern the photoresist. The photoresist is exposed to the interference pattern for 35 seconds. After the exposure is completed, the sample is placed in a developer for 1 minute, followed by DI water rinse for 1 minute (Fig. 2b).
- (5) The sample is placed in the reactive ion etching inductively coupled plasma (RIE/ICP) tool and the following recipe is executed:
The bottom plate of the RIE chamber is heated at 70° C. For the procedure described in Table 1 , step 1 is used to purge the chamber from any residual gasses; step 2 is used to transfer the pattern from the photoresist to the SiO2 surface (Fig. 2c); step 3 is used to etch the AlO2 followed by etching of the Al in step 4 (Fig. 2d). At the end of the procedure the chamber is purged in order to remove any residual chemicals that might be accumulated on the surface of the sample during the etching procedure. At the end of this step, the glass slide contains uniform aluminum nanowires across a circular area of 3 cm.
- (6) A negative SU-8 2002 photoresist is applied next on the sample. The negative photoresist, SU-8 2002, is hydrophobic and requires the applied surface to be absolutely free of any water molecules. The photoresist is spin coated at 500 rpm for 10 seconds and then at 3000 rpm for 50 seconds with 500 rpm per second acceleration. The resulting photoresist thickness is 2 μm. It is important to have a very good precise ramp up from 500 to 3000 rpm in order to obtain thickness of 2 μm (Fig. 2e).
- (7) The sample is baked at 65° C for 1 min and then at 95° C for 2 min on a hot plate. It is recommended that the sample cools down at 65° C for 1 min in order to gradually decrease the temperature of the sample. Gradual increase and decrease of the temperature during the baking process avoids rapid temperature differences and prevents the photoresist from cracking.
- (8) The photoresist is exposed at 375 nm wavelength for 22 seconds at 5mW/cm2 intensity using a Karl Zuess mask aligner. The mask used to pattern the photoresist contains 18 μm by 18 μm square checkerboard patterns (Fig. 2e). The sample is post-baked at 65° C for 1 min and then at 95° C for 3 min. The sample is cooled down at 65° C for 1 min in order to gradually decrease the temperature and minimize stress and cracking on the photoresist. The photoresist is developed for 3 min in an SU-8 developer using an ultrasound bath and gently rinsed with isopropyl alcohol at the end of the procedure. If white colored liquid appears on the surface, the photoresist is not completely developed and it is submerged in the developer again (Fig. 2f).
- (9) The sample is placed in the RIE/ICP tool and the recipe described in Table 2 Table 2 is very similar to the one in Table 1 and the last row in the table describes the purpose of each step. At the end of the RIE/ICP procedure, the glass slide contains pixelated nanowires polarization filters oriented at 0°. The pixel pitch of the polarization filters is 18 μm by 18 μm and the individual pixelated filters are composed of aluminum nanowires with 140nm height, 140nm periodicity and ~50% duty cycle (Fig. 2g).
- (10) The next step is to generate a second layer of pixelated nanowire polarization filters oriented at 45° offset from the first layer filters. First, a 250nm thick layer of SiO2 is deposited on the filter slide via PECVD. The SiO2 layer is polished and planerized via a CMP procedure. The thickness of the SiO2 layer is reduced by ~150 nm, such that the start of the second layer is immediately after the first layer of the aluminum nanowires (Fig. 2h).
- (11) Steps 2 through 10 are repeated in order to generate a second layer of pixelated aluminum nanowires polarization filters and are illustrate in Fig. 2i through Fig. 2o. In order to fabricate nanowires in the second tier whose orientation is offset by 45° with respect to the first layer, the glass sample is physically rotated by 45° prior to the interference lithography exposure described in step (4), i.e. Figure 2j. Also the same chromium mask with 18 μm by 18 μm openings is used as in step (9) and the mask is physically shifter by one pixel to the left and one pixel down compared to the first layer alignment in order to generate the necessary checker board pattern of pixelated polarization filters.
4. Results and measurements
The dual layer pixelated polarization filter is first visual evaluated under an optical microscope. The filter sample is backside illuminated and images are recorded under an optical microscope. An adjustable polarization filter is placed between the dual layer filter and the light source. The dual-layer filter is illuminated with 90° linearly polarized light and a microscope image is presented in Fig. 3a .
The dark square patterns in Fig. 3a are part of the 0° polarization filter layer, which are illuminated with orthogonal polarized light. The lighter square patterns are the 45° polarization filter layer structures and they are partially filtering the 0° polarized light. The transparent patterns in the image do not contain any polarization filters and the unfiltered light wave is passed through. In Fig. 3c, the sample is illuminated with 0° polarization light. The filter structures in the 0° polarization layer are transparent, while the filter structures in the 45° polarization layer have the same intensity as when the sample is illuminated with 90° polarization light. The pixels that do not contain any polarization filter do not change the intensity in both experiments. In Fig. 3b and Fig. 3d the filter array is illuminated with 45° and 135° polarized light and the polarization pixelated filters with nanowires oriented at 45° are transparent and opaque respectively. The pitch of the pixelated polarization filters in both layers is 18 μm.
The dual layer polarization filter is aligned and abutted against the surface of a custom made CMOS imaging sensor  via a 6-axis micromanipulator stage. The micromanipulator stage is designed using three translation stages, a rotating stage and two goniometers. The CMOS imaging sensor is composed of an array of 20 by 20 pixels with 18 μm by 18 μm pixel pitch. The optoelectronic performance of the CMOS imaging sensor is tested before it was merged with the polarization filters and the sensor has a dynamic range of 55dB, fixed pattern noise of 0.1% from saturation level and maximum signal to noise ratio of 43.3dB . The analog video signal from the CMOS imaging sensor is digitized via a 12 bit analog to digital converter and the digital information is transmitted to a PC via USB board from Opalkelly .
The optical performance of the dual layer polarization filter array is evaluated at three different wavelengths. A 2” integrating sphere with two ports is used to produce uniform light intensity. A halogen light source coupled via optical fibers provides input to the integrating sphere. A spectral filter is places at the output of the integrating sphere, followed by a linear polarization filter mounted on a computer controlled rotating stage. The angle of linear polarization is swept from 0° to 180° in steps of 1° and for every angle of polarization a total of 10k images are collected with the CMOS imaging sensor. The 10k images are averaged in order to increase the signal to noise ratio of the final image by a factor of 100.
Figure 4 illustrates the measured photo response of two pixels from the integrated CMOS imaging sensor with the array of pixelated polarization filters. The two neighboring pixels are covered with two different polarization filters offset by 45°.
The two polarization filtered pixels obey Malus’s law for polarization irradiance and are offset by 45° due to the physical orientations of the aluminum nanowires. Hence, the maximum and minimum transmissions between the two pixels are shifted by 45°. The maximum transmission for the 0° and 45° polarization pixels occurs when parallel polarized light illuminates the photodiodes and corresponds to 71% and 73% of the total incident intensity respectively. The minimum transmission is 0.89% of the total incident intensity for both pixels and is recorded when cross polarized light illuminates the pixel.
The photo response of the pixelated filters is evaluated as a function of the integration time of the imaging sensor. The photo response of the 0° pixel as a function of the integration time is presented in Fig. 5 . The integration time of the imaging sensor is varied from 1 msec to 43 msec in increments of 7 msec. For each integration time, the angle of polarization is swept from 0° to 180° in steps of 1° and 10K images are averaged to obtain a single reading. As the integration time is increased, more photons are collected by the underlying photo detector for both cross and parallel polarized light. Hence, both minimum and maximum values of the pixel are increased as the integration time is increased and can be observed in Fig. 5. However, the ratio of the maximum to the minimum value of the photo response remains constant for different integration periods i.e. different number of collected photons by the photodiode.
The extinction ratios for both pixels with 0° and 45° linear polarization filters as a function of integration time of the CMOS sensor are presented in Fig. 6 . The extinction ratios are computed as a fraction of the parallel polarized light, i.e. the maximum photo response, to the cross polarized light response, i.e. the minimum photo response. As the integration time is increased from 1 msec to 43 msec, the extinction ratios for the 0° and 45° pixel remain relatively constant at 80 and 82 respectively. The polarization sensor can accurately detect polarization information over a large range of collected photons, starting from minimum of ~100 photons up to full well capacity of the photodiode of 16K photons.
The difference in extinction ratios between the two pixels is primarily due to variations in the thickness of the nanowires. For example, the thickness of the 0° polarization filter is 135nm and the thickness of the 45° polarization filter is 142nm as measured via ellipsometry. These thickness variations are introduced during the chemical vapor deposition step of the microfabrication procedure.
The extinction ratios of the polarization filter arrays are evaluated at several different wavelengths. The maximum extinction ratio is achieved at longer wavelength, i.e. 700nm, and is 95. The minimum extinction ratio is achieved at shorter wavelength, i.e. 470nm, and is 51. The wavelength dependence of the nanowire polarization filters is an inherent property of the filters. If the width of the aluminum nanowires is decreased, the extinction ratios of the filter array can be improved as suggested by theoretical modeling .
We have presented a procedure for fabricating a micropolarization filter array suitable for realization of compact division of focal plane polarization imaging sensors. The fabrication procedure is composed of combination of interference lithography and regular microfabrication steps in order to realize a dual layer pixelated polarization filter array. The filter array is temporarily aligned against the surface of a custom CMOS imaging sensor and optical test indicate that the maximum extinction ratios of the filter array is 95 at 700nm. The fabrication procedure can be directly implemented on the surface of a CMOS or CCD imaging senor and would allow for monolithic integration of imaging and filtering on the same substrate.
This research is supported by Air Force Office of Scientific Research grant numbers FA9550-10-1-0121 and the CMOS chips were fabricated through MOSIS.
References and links
1. J. Wang, F. Walters, X. Liu, P. Sciortino, and X. Deng, “High-performance, large area, deep ultraviolet to infrared polarizers based on 40 nm line/78 nm space nanowire grids,” Appl. Phys. Lett. 90, 061104.1–061104.3 (2007).
3. J. G. Ok, H. J. Park, M. K. Kwak, C. A. Pina-Hernandez, S. H. Ahn, and L. J. Guo, “Continuous patterning of nanogratings by nanochannel-guided lithography on liquid resists,” Adv. Mater. (Deerfield Beach Fla.) 23(38), 4444–4448 (2011). [CrossRef] [PubMed]
8. T. Tokuda, S. Sato, H. Yamada, K. Sasagawa, and J. Ohta, “Polarisation-analysing CMOS photosensor with monolithically embedded wire grid polarizer,” Electron. Lett. 45(4), 228–230 (2009). [CrossRef]
9. X. Zhao, F. Boussaid, A. Bermak, and V. G. Chigrinov, “Thin Photo-Patterned Micropolarizer Array for CMOS Image Sensors,” IEEE Photon. Technol. Lett. 21(12), 805–807 (2009). [CrossRef]
10. B. E. Bayer, Color Imaging Array U.S. Patent 3,971,065, (July 20, 1976).
11. G. P. Nordin, J. T. Meier, P. C. Deguzman, and M. W. Jones, “Micropolarizer array for infrared imaging polarimetry,” J. Opt. Soc. Am. A 16(5), 1168–1174 (1999). [CrossRef]
12. S. Franssila, Introduction to Microfabrication (John Wiley & Sons, West Sussex, UK, 2010).
13. J. S. Tyo, “Optimum linear combination strategy for an N-channel polarization-sensitive imaging or vision system,” J. Opt. Soc. Am. A 15(2), 359–366 (1998). [CrossRef]
14. M. L. Schattenburg, R. J. Aucoin, and R. C. Fleming, “Optically matched trilevel resist process for nanostructure fabrication,” J. Vac. Sci. Technol. B 13(6), 3007–3011 (1995). [CrossRef]
15. V. Gruev, Z. Yang, J. Van der Spiegel, and R. Etienne-Cummings, “Current mode image sensor with two transistors per pixel,” IEEE Trans. Circuits Syst. I: Fundam. Theory Appl. 57(6), 1154–1165 (2010). [CrossRef]